interview

When looking for a summer internship opportunities, I keep in mind the two basic standards.First of all, I'd like to work in the fields of deep learning.Second, I really want to practice somewhere in Japan.

In the past year, as a sophomore, machine learning, especially the deep learning has always been my spontaneous learning efforts.Through MOOC, individual, voluntary research assistant and internship, I have laid a substantial knowledge base on these subjects, so far, these are not my course content.In order to further improve my skills, I feel it is necessary to obtain scientific questions about my perception data and learn the way to solve the problem of feedback by depth.I will focus on looking for internship search in the role of a deep learning as the center.

In addition to the subjects of the internship, internship site in Japan the situation as well as its short term and long term interest is a decisive factor for me.I worked in studied Japanese Osaka university and got the MEXT scholarship, so I'm looking forward to with Japan and Japanese.I also want to experience life as r&d engineer in Japan.

Due to meet the two conditions, open data analysis in SEI internship is the perfect choice for me.

During the internship, I implemented a deep learning model, this model can be given a set of objects of classes (roads, buildings, vegetation, etc.) to satellite image segmentation.In other words, this corresponds to the construction of artificial neural network, the artificial neural network can on satellite images of a set of tags after the training, learn to tag on any of the satellite image mentioned categories.I chose a kind of widely used in image segmentation of neural network architecture, the architecture is called "net", and through the training, validation, and fine-tune cycle to improve the prediction precision.After get the final result from the test data, I will present results to data analysis, the department manager and my team of engineers and hr specialist.

This may let a person feel a bit strange, but "public stretch time" benefit from the development of the office, I found that stretch after lunch to rest for a few minutes can actually make the whole afternoon is full of vitality.Although at first most of my colleagues carefully hips and knead shoulder feel some doubts, but I later realized truly awakening and take good care of the body have a magical effect on improve the work efficiency.I feel I have formed the habit of strong, now it is suitable for the need of mental work situations.

First of all, this practice improved my work in the data analysis.I really like in machine learning program, especially this brought me into the project in the field of computer vision.This will, of course, for my future career focus and research interest in a lasting impact.

In addition to do the practical work in the internship program, had the opportunity to experience the world first-class companies r&d environment makes me engaged in r&d career expectations closer to reality.It helps me to better determine the field of professional what's good for me.

In addition, this attachment is my first time to go abroad to work.Easy communication with company everyone and support for interns encountered various problems left a positive impression on us.I have been considering to work or study abroad after graduation, and I work abroad for the first time in SEI pleasant experience I've definitely got strengthened that resolve.

If you decide to work in the company's full-time, you should be keen to identify which projects you want to do.Your internship program will be a good indicator.But the other team members (especially your mentor) is engaged in the project is to measure the question you might encounter in the study of the scale and complexity of good method.Actively interact with the teammate, in order to better understand the work they do.Your daily work may don't need you to cooperate with a person other than the teacher, but SEI, everyone is happy to provide guidance and detailed explanation when asked.